EACON Mining Technology, a leader in autonomous mining haulage, has been awarded “best method” for its interactive prediction algorithm – mixed-traffic environment in the 2025 Argoverse challenge.
Presented at the Computer Vision and Pattern Recognition (CVPR) conference workshop on autonomous driving. The event, held in Nashville, Tennessee, brought together leading researchers and engineers from academia and industry to showcase the latest advancements in autonomous driving.
Competing against more than 100 international submissions, including top-performing teams from previous years, EACON achieved the highest score in the argoverse 2 multi-agent motion forecasting task.
With more than 1800 trucks operating across 24 sites, the company has first-hand experience of the unique prediction challenges mining presents compared with road-based autonomous driving.
Manually driven mining vehicles move unpredictably – overtaking, avoiding obstacles, or altering speeds, creating high uncertainty for autonomous systems.
EACON’s award-winning IMR (iterative mode-world weighted regression) framework addresses these challenges by accurately distinguishing between behavioural patterns and establishing clear movement boundaries.
This enables autonomous mining trucks to predict with greater accuracy, improve the smoothness of mixed-fleet interactions, enhance safety by reducing collision risk, and boost efficiency through streamlined haul road traffic flows.
The IMR framework is already delivering results in operational mining environments. At the Yihua Mine in northwest China, traffic time at a mixed intersection has been reduced by 20 per cent, enabling smoother and safer interactions between autonomous and manual vehicles.
The technology has also reduced unnecessary braking in autonomous haulage, with heavy braking down 10 per cent, medium braking down 43 per cent, and light braking down 50 per cent per km – significantly reducing vehicle wear and tear.
EACON has led the Argoverse2 leaderboard since August 2024 with its self-developed joint trajectory prediction framework.
“Our IMR framework is a milestone in mining autonomy. It allows our autonomous haulage systems to operate more safely and efficiently alongside manned fleets in complex, unpredictable conditions,” EACON Mining Technology algorithm lead Tristan Liu said.
“This recognition at CVPR is not just about research excellence, it’s proof that our solutions deliver measurable operational gains in real mines today.”
EACON has also recently launched its first Australian trial in partnership with Thiess and Norton Gold Fields. The retrofit of a Komatsu HD1500 haul truck with its proven OEM-agnostic autonomy has been completed, and the groundbreaking autonomous solution is now undergoing live trials in Kalgoorlie.
Want to connect with the mining industry? Register to attend AIMEX and WA Mining Conference.
